Smart Machines

����������� Intelligence is something that has many qualities to it, and no real definition. For our purposes, we can define intelligence based on what it needs to be for what we need it to do. It seems that a machine can be intelligent if it meets a few basic requirements. Our basic qualities would be independent data output and the ability to learn in a beneficial manner. Learning can come from its own outputs, or from an outside source feeding it data. A machine has artificial intelligence if it�s able to comprehend the inputs it receives and appropriately output data that corresponds to the inputs. It is able to output this data independently without set rules telling it what to do. It also has to act independently without set code that says �if this happens do this�. The machine also has to be able to interact on imperfect information systems. Because if the system is a perfect information system, there is no real intelligence happening. All data present can lead to an appropriate result without any �thinking� or �intelligence� involved. A good example of this is chess. At each turn, all information is known, and the machine can output a move which is the possibly the best move it can make. With imperfect information systems, the ability of the machine to make �intelligent� decisions deals with the intelligence decision making ability that it has. Such a game would be Poker. Since you have no clue what cards your opponents have, making a decision based on what you have and how they bet is the only way that it can be done. Learning is a vital part of a smart machine. If a machine can�t learn, then it can�t possibly have artificial intelligence. If learning is not involved, then there can�t ever be an intelligent decision made because the machine would just be �randomly� making a decision without any internal �intelligent� decision making procedures. Even with a poor learning system within the machine, it should be able to make slightly better decisions each time. With this learning idea, we can apply it to other things besides poker. On the other end, what happens if the machine learns incorrectly? Can we say that this is still a smart machine? Since the learning is not beneficial to the situation, it can�t be considered intelligent.

There are also situations where the data outputted doesn�t really have a right or wrong answer, because the data contains something of human qualities. An example of this is the EMI machine which outputted music. How can a computer tell if something sounds good or not? Or perhaps one type of sound some people like an others dislike. Or with the EMI machine, if it produced music that didn�t sound good, could it still be considered a smart machine? I believe that since it�s taking in input, and outputting something new that it produces, that it still is a smart machine.